This paper explores the possibility of using approximate model for fitness landscape in multi-objective optimization. A multiobjective genetic algorithm based optimizer, namely, NSGA-II is integrated with artificial neural network (ANN). This presented technique makes use of successive fitness landscape modeling for reducing the precise function evaluation calls while retaining the basic robust search capability of genetic algorithms (GA). The procedure is tried on some of the standard test problems available in literature on multi-objective optimization. The simulation results show a considerable savings in precise function evaluations and a good diversity in the obtained Pareto-front.